dr-strangecode / vagrant-centos-rackspace-workaround

workaround to CentOS woes with vagrant - specifically for rackspace

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Vagrant Centos / Rackspace Workaround

Vagrant will not work properly with CentOS out of the box. This has to do with how vagrant handles ssh sessions and how the sudoers file is configured by default on CentOS.

This isssue is confirmed on CentOS and assumed to affect all RHEL variants.

The "worst" part is that we must do this for every dc we wish to deploy to with vagrant.

NOTE: There is currently a problem with the vagrant-rackspace plugin. Please refer to Pull Request 11 for details. This fix is required to be able to deploy to ORD with vagrant. If you are only interested in DFW, then you can skip this step.

NOTE: There is also an issue with LON. A bit of a dirty workaround can be seen in Pull Request 4

Usage

  • ensure vagrant is installed via package: Vagrant Downloads
  • vagrant plugin install vagrant-rackspace
  • git clone git@github.rackspace.com:lefty/vagrant-centos-rackspace-workaround.git
  • cd vagrant-centos-rackspace-workaround
  • vagrant box add dummy https://github.com/mitchellh/vagrant-rackspace/raw/master/dummy.box
  • edit Vagrantfile adding in your username and api key
  • vagrant up --provider=rackspace for each machine. e.g. vagrant up temp-centos-6-dfw --provider=rackspace
  • use vagrant to ssh in to each machine. e.g. vagrant ssh temp-centos-6-dfw
  • visudo
  • comment out Defaults requiretty
  • comment out Defaults !visiblepw
  • save file and logout of machine
  • from the control panel, create an image of that machine with a name like "vagrant-centos-6-dfw"

Once images have been created, you can start using vagrant normally with those saved images. Don't forget to remove the temporary machines.

Error Details

Error:

The following SSH command responded with a non-zero exit status.
Vagrant assumes that this means the command failed!

echo $(chef-solo --v | awk "{print \$2}" || "")

The only workaround at this time is to create a custom image with sudoers configured to work with vagrant.

Fix:

  • create machine manually or with vagrant (suggest a small instance)
  • ssh into machine as root and edit /etc/sudoers with visudo
  • comment out Defaults requiretty
  • comment out Defaults !visiblepw
  • from the control panel, create an image of that machine
  • update Vagrantfile to use the new image instead of the "normal" one

Note: If you need to create CentOS / RHEL machines in multiple dc's, you will need to do this for every dc.

About

workaround to CentOS woes with vagrant - specifically for rackspace


Languages

Language:Ruby 100.0%